My colleague had an interesting question

Why should we do showcases if no one will action our work?

Their point was fairly valid. We had set up a git repository with our work, sample code to get started on a variety of “difficult” problems which people across the organisation have generally been interested in. However, the consensus within the team was that no one outside of the team had actually cloned the repository; much less tried running any of the code.

Knowledge sharing, showcases, lunch and learn, brown bag sessions; these things have many names. They all fall under the heading of “knowledge sharing” which is commonly associated with agile and a with the goal to grow communities. Ultimately though; if you work in an environment where the things you do are fundamentally different to other people, the overlap and opportunity to try our new knowledge just isn’t there.

Which of course begs the question,

Why?

Why would you do this? Personally there are three core reasons why - regardless whether people actually do learn (or not) from my knowledge share sessions, I will still continue to run them. Initially many of these sessions were just run with the immediate team, however over time more and more people have been interested and it has now extended to the wider analytics community.

Maintaining Relationships

In my experience there are very limited avenues for different analytics teams within the same organisation to share ideas. Often times this is not due to the lack of trying, but simply because there is little incentive to do so. Knowledge sessions are a great way to connect with people who you might not otherwise have met. They could be in another team or even another city. There have been several new contacts which I have made this way, and I can say that I have helped improved their own analytical workflow through introducing them to my own knowledge sharing sessions.

However when we expand knowledge sessions with the wider communities, the practical challenges associated with operation analytics is often times different to customer or marketing analytics, and again usually different to anything relating to pricing or credit risk. Although in many cases you might use similar theoretical models; most people’s day to day work is not done in theory!

Nevertheless it is important to discuss a new model, a new approach, a new process - after all you might just help someone else in the organisation, in the process of helping yourself!

For Your Own Improvement

A common comment which I hear often about analytics professionals is that analysts are poor public speakers, but excellent technicians. Personally I find that comment to be ridiculous!

From the start of your career in data analytics, the expectation is that you spend every week, week in and week out working on models, designing reports and providing insights. One would probably spend the whole week (say 30+ hours) working on this. Perhaps, if you’re lucky, once every two months you are asked to present your findings in a 30 minute timeslot. Most people wouldn’t spend much more than 30 minutes preparing for this presentation.

If we consider the breakdown above, in two months an analyst would have spent 8 * 30 hours = 240 hours on analytics and 1 hour on public speaking. Of course they will be fantastic in quantitative analysis and poor at public speaking! Just like how professional athletes are expected to train in order to be peak form during a competition, analysts also should train in order to perform during presentations. Knowledge sharing sessions are the perfect avenue for this.

If you can’t explain it simply, you don’t understand it well enough

There is of course the teach component of knowledge sharing. When you are asked to teach someone, you are forced to consider every single step or process which you needed to follow to get the end result. In knowledge sharing sessions, you are now forced to explain what you know in terms which a broad audience can understand.

This aspect of it acts like the 2nd opportunity to learn a topic, and ensures that you can explain it to anyone else in the future; because they might take something you said on board and try it for themselves.

Inspire Others

You give a poor man a fish and you feed him for a day. You teach him to fish and you feed him for a lifetime

As a steward of the data science team, perhaps an unwritten rule in our job description is to get other people excited about big data. They say that there is a massive knowledge gap, where we need more and more people with great programming expertise and quantitative knowledge.

Unfortunately, the number of people who can mentor and upskill others are far and few. In my opinion it is in our best interest to motivate and inspire others to learn for themselves. That is what knowledge sessions can offer! You could demonstrate the power of word2vec or LDA algorithms and how to go beyond simple bag-of-word models for text mining, get people excited about the possibilities of deep learning and image recognition, explain the process and power of ensembles and how you use them to improve your predictive models whilst in production!

I thank my colleague for asking me this question, because it was a genuine question which got me thinking, as we might sometimes wonder why is it we do things. Like many aspects of our working life, sometimes we might be ecstatic over the success of a project, other times we might get discouraged, wondering “why are we even doing this at all?!”

We should always refer back to our core principles to why we act and do certain activities - like this knowledge share.

Afterall, success is not always what you see.